Design system behavior to preserve human agency.
Ensure automation supports rather than replaces user decision-making.
Contents 4. System Behavior & Workflow Logic 3. Autonomy
How much control users retain.
First, ask yourself
“Do users of your system retain meaningful control over consequential actions?”
Mission statement
Ensure users delegate safely by enabling controlled system actions without loss of oversight and control.
Key heuristics Autonomy
- System behavior must support, not replace, user decision-making.
- Consequential automation must require user awareness and consent.
- Users must be able to override, pause, or modify automated processes.
- Automated activity and intent must be perceptible.
- System-initiated actions must be undoable or revisable.
- User control requirements must increase with consequence severity.
- Automation must not be enforced solely for efficiency gains.
- Users must be able to inspect and adjust persistent automation and history.
- Automation must strengthen users’ sense of control and capability.
Executive brief
The system must preserve real user control over delegated actions.
Core questions Autonomy
“Which decisions are made by the user, and which by the system?”
“Can users see, understand, and control delegation?”
Focus areas
Control-first
“Can users always override the system?”
“Is delegation deliberate or implicit?”
Risk-aware
“What happens if the system is wrong?”
“Are high-risk actions ever taken without consent?”
Transparency-aware
“Are system decisions and rationale visible to the user?”
“Can users predict when and why the system will act?”
AI-aware
“Is automation advisory or authoritative?”
“Does learning expand capability without reducing user control?”
UX directives Autonomy
Require explicit delegation of automated behavior.
Do not perform consequential automation without user awareness and consent.
Provide meaningful intervention mechanisms.
Enable users to override, pause, or modify automated processes.
Make automated state and intent visible.
Ensure users understand what the system is doing and why.
Ensure reversibility of automated outcomes.
Allow users to undo or revise system-initiated actions.
Calibrate autonomy according to risk level.
Increase user control requirements as consequence severity rises.
Prioritize agency over raw efficiency.
Avoid enforcing automation solely for speed gains.
Preserve autonomy across time horizons.
Enable users to inspect and adjust persistent settings and historical automation.
Design automation to reinforce user confidence.
Ensure system behavior strengthens users’ sense of control and capability.
Executive summary
- Autonomy is preserved human agency, not unrestricted automation.
- It ensures the system supports decisions without replacing them.
- The system must require explicit delegation for consequential automated behavior.
- Automated state, intent, and outcomes must remain visible, controllable, and reversible.
- Autonomy must be calibrated to risk, increasing user control as consequence severity rises.
- Autonomy succeeds when automation strengthens users’ sense of control rather than diminishing it.
Success indicators
- Users remain in control of important decisions.
- Automated actions occur with user awareness or consent.
- Users can pause, modify, or override automation.
- The system clearly shows what automation is doing.
- Automated actions can be undone or corrected.
One-line summary
Before users can trust a system, they must feel it acts with them, not for them.